In today's world, making millions of data understandable has become important. To take faster steps in criminal matters, especially by using these data, data analysis should be done quickly. In this context, sentiment analysis performed with the natural language processing (NLP) method of artificial intelligence enables the elimination of possible loss of life and property. In addition, by listening to all radio frequencies at the same time in possible terror areas, the attacks of terror organizations can be analyzed with natural language processing methods, so that the attack can be prevented before it takes place. In this study, natural language processing methods of artificial intelligence were used in the analysis of text, audio, and image data in the virtual environment for the detection of terror threat elements. In this way, it is aimed to ensure the healthy intervention of law enforcement officers and the security of life by analyzing the talks of terror elements in terror zones. For this purpose, an 85% accuracy rate was reached with the word/sentence vector creation method GloVe in the first model created with the Spark NLP library on textual data. In addition, a 74% accuracy rate was achieved with the LSTM method on audio data, while a 71% accuracy rate was achieved with the GRU method on visual data.
deep learning natural language processing data preprocessing feature extraction classification methods terror-threat elements
Primary Language | English |
---|---|
Subjects | Natural Language Processing |
Journal Section | Research Articles |
Authors | |
Publication Date | August 31, 2024 |
Submission Date | July 17, 2024 |
Acceptance Date | August 20, 2024 |
Published in Issue | Year 2024 Issue: 010 |